Quality of Service Specification for Resource Management in Multimedia Systems
Total Page：16
File Type：pdf, Size：1020Kb
Quality of Service Specification for Resource Management in Multimedia Systems Richard Alan Staehli B.S., The Evergreen State College, 1982 A dissertation submitted to the faculty of the Oregon Graduate Institute of Science & Technology in partial fulfillment of the requirements for the degree Doctor of Philosophy in Computer Science and Engineering January 1996 @ Copyright 1996 by Richard Alan Staehli All Rights Reserved The dissertation "Quality of Service Specification for Resource Management in Multi- media Systems" by Richard Alan Staehli has been examined and approved by the following Examination Committee: - . Jonathan v3"aIpo$ Associate Professor Thesis Research Adviser David Maier Professor johnPrincipaNvl Member Tech. Staff GTE ~aboratories,Inc. iii Dedication To my wife, who yearns for me to become a gardener. Acknowledgements The most important person to acknowledge is my advisor Jonathan Walpole. Without his encouragement and advice, I would not have entered the Ph.D. program at OGI. I have always felt strongly that taking time to hike, to climb, to ski and sail are as essential as any other pursuit in this life. Jonathan demonstrated that one can continue outdoor pursuits and still pursue science with a dedication that I had previously associated with digging a snow cave in a blizzard. In addition, many of the ideas in this thesis evolved slowly from our talks so that it is impossible to claim sole credit. Both Jonathan and Dave Maier have provided suggestions and feedback, beginning with my research proficiency paper five years ago and continuing up to our most recent publication. While this thesis represents work that I have developed and written on my own, I gratefully acknowledge the valuable criticism that Jonathan and Dave have provided. I would also like to thank the students that have formed my at work family for these past years. Ravi Konuru and Harini Srinivasen were good friends through our first year of classes and the qualifying exams. They each inspired me by finishing their thesis work in good time and with good results. The other members of my "quals group" include Judy Cushing, Steve Rehfuss, Jon Inouye, Moira Mallison, and Bennet Vance. All have been valued friends and colleagues for sharing and learning the tricks of Ph.D. candidacy. I owe Judy a unique debt for having introduced me to computer programming when I took her Data Structures course at The Evergreen State College in 1982. Her letter of recom- mendation was probably a key to landing my first job out of college! Her creative energy continues to provide a fountain of opportunities for research and professional collabora- tion. I'd also like to acknowledge the many times that Bennet listened to my description of a problem that I could neither solve nor articulate well. It proved useful to know what parts of the problem I could communicate to him. In addition to these close friends that took the qualifying exams with me, I need to mention Scott Daniels, who was like a big brother to me, Jenny Orr, and Nanda Kambhatla, who have always formed the core of the party crowd in CSE, such as it was. I regret that we didn't all have more time together for recreation. I've deliberately delayed acknowledging the assistance of Jon Inouye. Of course, I need to thank Jon for the official BwOGI Ph.D. thesis template. This is only one of many examples where Jon was so capable and willing to provide expert consultation on tools that I used without studying. I knew from previous industrial experience that an organization works best if it cultivates an expert or two for each tool. These experts take responsibility for maintaining and upgrading the tool and provide intelligent tutoring to other tool users. Jon Inouye has provided this expert service for B',HPUX, Mosaic and many other tools that have been essential in writing my thesis. His sushi-making skills and provisioning of the lab cookie jar are also much appreciated. Other members of the faculty that have been of particular help include Jim Hook who taught a mean Automata Theory course in his very first quarter of teaching at OGI, and David Novick who was my first mentor and also advised me to enter the Ph.D. program. Harry Porter, a graduate student at the time, introduced me to object-oriented programming in his 1985 OOP class. Roger Lea, while only an adjunct faculty member for a brief time, gave me some of the most memorable advice about completing my Ph.D. work and its only through my own stubborn nature that I was unable to follow it well. I am also very grateful to Tom Little of Boston University for inviting me to meet with his research group in May of 1994 and to John Nicol of GTE Labs for inviting me to speak there during the same trip. These groups endured a painfully early version of my thesis talk. Of course, I am very grateful to the members of my thesis committee for each giving this thesis a careful reading. Financial support for this work came from National Science Foundation grant IRI- 9117008 for Multimedia Storage research and IRI-9223788 for Scientific Database research. Additional support came from Tektronix, Inc., the Portland Trailblazers, and from AT&T Bell Labs. I would particularly like to acknowledge Bart Locanthi from Bell Labs who built and loaned us our first digital video cameras. My first programs for capturing and manipulating video images were executed on a MIPS Magnum running the Plan 9 operating system that Bart had installed and loaned to us. I deeply regret that we were not able to make a commitment to use Plan 9 for our other work. The opportunity to work with such a well crafted and documented set of tools was a great joy. Special thanks to my wife, Jan, for providing both pressure to complete and respite from the task of writing this thesis. We have acted together to see this goal achieved, not knowing how it would change our lives. vii Contents Dedication ..................................... iv Acknowledgements ................................. v List of Tables .................................... xi List of Figures ................................... xii Abstract .......................................xiv 1 Introduction 1 1.1 Multimedia Systems and Resource Management ................ 1 1.2 The Need for Presentation Quality of Service Specifications ......... 5 1.3 Scope and Contribution of this Thesis ..................... 8 1.4 Outline of the Thesis ............................... 9 2 Real-Time Presentation of Stored Multimedia Data 10 2.1 Identifying Presentation Requirements ..................... 10 2.1.1 Continuous Media ............................ 11 2.1.2 Muse .................................... 12 2.1.3 Object Composition Petri Nets ..................... 13 2.1.4 MAEstro ................................. 14 2.1.5 Algebraic Video ............................. 15 2.1.6 MHEG .................................. 15 2.1.7 Discussion ................................. 17 2.2 Presentation Techniques ............................. 18 2.2.1 Output Computation .......................... 19 2.2.2 Synchronization ............................. 20 2.2.3 Storage Optimizations .......................... 24 2.2.4 Process Optimizations .......................... 27 2.2.5 Data Compression ............................ 28 2.2.6 Prefetching ................................ 30 viii 2.2.7 Resource Reservations .......................... 34 2.3 Summary ..................................... 37 3 Reference Architecture 40 3.1 Reference Architecture Description ....................... 40 3.2 Application Examples ..............................42 3.2.1 Electronic News Gathering ....................... 42 3.2.2 Digital Video Support for Professional Sports ............. 46 3.3 Summary .....................................48 4 Specification of Presentation Quality 49 4.1 Z Notation .................................... 50 4.2 ' Content Specification ............................... 51 4.2.1 Content Descriptors ........................... 52 4.2.2 Semantics ................................. 55 4.3 View Specification ................................ 60 4.3.1 View Descriptors ............................. 61 4.3.2 Semantics ................................. 63 4.3.3 Actual Presentation ........................... 64 4.4 Quality Specification ............................... 65 4.4.1 Reference Error Model .......................... 70 4.4.2 Quality Descriptors ........................... 74 4.4.3 Semantics ................................. 75 4.5 Summary ..................................... 77 5 A QOS-Driven Multimedia Player 79 5.1 Purpose and Scope of the Prototype ...................... 79 5.2 QOS Request Generation ............................ 82 5.2.1 Creating and Selecting Content ..................... 82 5.2.2 View Controls .............................. 83 5.2.3 Quality Calibration and Constraint ................... 86 5.2.4 Representation of QOS Requirements ................. 88 5.3 Presentation Planning .............................. 88 5.3.1 Components of a Presentation Plan .................. 89 5.3.2 Admission Test .............................. 91 5.3.3 Proof of QOS Guarantees ........................ 96 5.4 Presentation Execution .............................100 5.4.1 Resource Overload Detection and Handling ..............100 5.5 Discussion .....................................101 6 Related Work 103 6.1 QOS Specification ................................103 6.2 Presentation Planning ..............................106 7 Conclusions 108 7.1 A Framework for Defining Formal